Fix x86_64-unknown-illumos LLVM target triple

[madsmtm]

We were passing "pc" as the vendor component, which isn't present in the rustc triple, and is inconsistent with the Aarch64 target.

Motivation: To make it easier to verify that cc-rs' conversion from rustc to Clang/LLVM triples is correct.

CC target maintainers @jclulow and @pfmooney.
r? jieyouxu

[rustbot]

These commits modify compiler targets.
(See the Target Tier Policy.)

[jieyouxu]

Please r=me if target maintainers agree.
@bors delegate+ rollup

[bors]

✌️ @madsmtm, you can now approve this pull request!

If @jieyouxu told you to "r=me" after making some further change, please make that change, then do @bors r=@jieyouxu

[jieyouxu]

@rustbot blocked (waiting to hear back from target maintainers, not much for me or PR author to do)

[pfmooney]

We were passing "pc" as the vendor component, which isn't present in the rustc triple, and is inconsistent with the Aarch64 target.

It's true that those platforms do differ in their triples, but I'm not sure we can simply change the amd64 side. Have you built an illumos toolchain with this in place to check if the LLVM interactions are impacted?

[madsmtm]

We were passing "pc" as the vendor component, which isn't present in the rustc triple, and is inconsistent with the Aarch64 target.

It's true that those platforms do differ in their triples, but I'm not sure we can simply change the amd64 side. Have you built an illumos toolchain with this in place to check if the LLVM interactions are impacted?

Nope, was kinda hoping someone actually experienced with Illumos would ;).

I have looked at all usage sites of (Triple|TargetVendor)::(PC|UnknownVendor) in llvm-project though, the distinction is only used once in LLVM itself, and that's in a Mach-O specific code path. All other usage of these are in Clang and LLDB, so I'm fairly confident that it won't have any effect.